Hi Pauline

> Does anyone what is the default perturbation value in the LWC profile
> used to compute the Jacobians ?

There is no value to define as these Jacobians are calculated in an 
analytical manner. No numerical perturbation is performed.

Some comments here:
The above is throughout true for "abs species" (as long as you use the 
analytical version), but some internal hard-coded values could be used 
for other quantities even if they are "analytical".  I have tried to 
write semi-analytical to indicate this. For atmospheric variables, the 
radiative transfer part is throughout handled by analytical expressions, 
but the derivative of the absorption is done differently depending on 
ARTS version and the way absorption is obtained.

Let us use temperature as example. When using an absorption lookup 
table, the derivative of absorption with respect to temperature is done 
by an internal "perturbance" (with a hard-coded value). For ARTS v2.2 
the same is true with "on-the-fly" absorption, but in v2.3 this is now 
done in an analytic manner. This improvement was made by Richard 
Larsson, quite recently. Thanks for that!



> Is there any way to output the Jacobians without running a retrieval
> only by calling ARTS ? (like calling arts_y(Q) to output brightness
> temperatures).

Both yes and no. Note that arts_y can also return J. So if you set-up 
your Q in a way that it will be accepted by arts_y, the answer is yes. 
If you have managed to use arts_y, it should not be difficult to also 
get out J. Just make sure that you set J_DO=true.

Otherwise my suggestion is to run qpack but just doing a linear retrieval.

Bye,

Patrick




_______________________________________________
qpack mailing list
[email protected]
https://www.sat.ltu.se/mailman/listinfo/qpack

Reply via email to